Deficits in gestural joint-attention behaviors are a prominent feature of young autistic children. Attempts to explain these deficits have called upon the metarepresentational deficit hypothesis (Baron-Cohen, 1988; Leslie & Frith, 1988). However, developmental research suggests that joint-attention skills emerge prior to the cognitive capacity for metarepresentation. Thus, the metarepresentational hypothesis does not appear to provide a parsimonious explanation of autistic joint-attention deficits. An alternative model is proposed that attempts to explain these deficits in terms of the combined negative impact of developmental disturbances in affective, as well as cognitive, processes.

Mundy and Sigman (1989) present a challenge to the metarepresentational theory of autism. Their argument holds if it is assumed that the capacity for metarepresentation only emerges at 12–24 months old in normal development, as manifested in pretend play. However, I present reasons for postulating that the capacity for metarepresentation may be present from as early as 7–9 months of age, manifested in joint-attention behaviors. If this account has any validity, then the metarepresentational theory of autism is unaffected by Mundy and Sigman's challenge. In addition, I highlight some problems in Mundy and Sigman's alternative model of autism. These criticisms aside, their article is without doubt important and opens up critical questions both for models of the origins of autism and of a theory of mind in normal development.

I argue that Mundy and Sigman (1989) are right to question Leslie's (1987) explanation of the autistic syndrome, but that they do so for the wrong reasons. They assert that there is a lag between the emergence of joint-attention and pretence. However, evidence for the lag is equivocal because it depends on the behavioral criteria for each skill. Moreover, even if there were a lag, it could be argued that the relatively early appearance of joint-attention simply shows that a decoupling device is available before its recruitment for pretence. More problematic, with regard to Leslie's theory, is the finding that autistic children can engage in certain forms of pretence. This result suggests that autistic children suffer not from a general incapacity to conjure up imaginary entities or properties, but from a more specific impairment in the ability to imagine mental states that they are not currently experiencing.

Mundy and Sigman (1989) note that a problem arises in reconciling the existence of joint-attention deficits in autism with Leslie's (1987) theory that autistic children lack an innate, late-appearing metarepresentational capacity. Such a theory fails to encompass earlier rooted and perhaps more basic manifestations of autistic children's impairments in interpersonal coordination and intersubjective communication. The account provided by Mundy and Sigman, while promising, fails to address a critical question: How is it that autistic children have limited understanding that other people are subjects of experience with whom things can be shared? 1 suggest that an answer emerges only when we pay due regard to the nature of normal and autistic interpersonal relations. Here the authors' cognitive-behavioral analysis of affective exchanges between people leaves out the heart (and much of the mind) of the matter.

Recent empirical and theoretical work on both normal and pathological development has led to the formulation of the metarepresentational conjecture for autism. This account of autistic development links the core impairments in imaginative abilities, communicative skills, and social competence to dysfunction of a single set of underlying cognitive mechanisms. In this context, Mundy and Sigman (1989) argue that the impairment of early gestural communication in autism cannot be accounted for by the metarepresentational theory. On the contrary, this early communication impairment in autism is highly consistent with a faulty theory-of-mind module. In normal development, this module is available from around the first year to handle estensive communication – shared pretense being a prime, though slightly later, example. We briefly consider the role in communication of perceptual processes that may trigger metarepresentation. This early theory-of-mind module, with its metarepresentational processes, provides the specific developmental basis for later versions of the child's theory of mind.

This article presents our response to the comments of Baron-Cohen, Harris, Hobson, and Leslie and Happé. We suggest that a singular cognitive hypothesis does not provide a parsimonious explanation of autism. We argue that certain aspects of autism, including observations of joint-attention deficits and observations of deficits in the prosodic elements of speech, may best be explained in terms of both cognitive and affective factors. We also acknowledge the validity of the criticism of our contingency processing deficit hypothesis (Mundy & Sigman, 1989a). In response to this criticism, we offer a modification of our model of joint-attention skill deficits in autistic children.

A structured childhood history interview was administered to 50 low-income mothers of infants. Three levels of risk for infant maltreatment were represented in the sample, including 10 maltreatment cases. Maternal behavior at home was rated and factor analyzed when the infants were 12 months and 18 months old, yielding separate factors labelled maternal involvement and hostile-intrusiveness. Risk for maltreatment was strongly correlated with the overall adversity of the mother's childhood history. Maternal hostile-intrusive behavior was best accounted for by psychiatric disorder in the mother's own mother and poor peer relationships in childhood. Maternal involvement was negatively associated with a separate cluster of variables indexing family disruption and lack of supervision. The emergence of two separate clusters of associations — positive affective involvement and negative affective involvement — is related to similar two-dimensional results in studies of self-reported mood, infant attachment relationships, and peer relationships. Implications of a two-dimensional affective construct for future theorizing in developmental psychopathology are discussed.

This research examined the impact of child maltreatment and grade level on perceptions of competence and explored relations between perceived competence and school functioning among a sample of 104 low-income school children, half of whom belonged to families in which child maltreatment had been officially documented. Data were collected on 76 first through third graders and 28 fourth through sixth graders, who were interviewed in a laboratory setting, and on a subset of 74 children, who were evaluated by their teachers on measures of school functioning. Results indicated that maltreated children perceive themselves and/or describe themselves differently than do their low-income peers. However, these differences took the expected negative direction only beginning around the middle of the elementary school years. In earlier grades, maltreated children demonstrated a tendency to describe themselves in exaggerated positive terms, over and above the self-enhancement bias documented here and in other research among younger nonmaltreated children. Differences between maltreated and comparison children also appeared in teachers' ratings of school functioning and in use of special educational services. These results are presented and discussed as suggestive evidence for a causal chain from parent-child relationships, to self-perceptions, to school functioning.

Adolescents attempt to end their own lives with greater frequency than do either younger or older persons. The aim of this study was to provide a developmental account of this anomaly by examining the contrastive ways in which suicidal and nonsuicidal adolescents reason about their own personal continuity through time. Drawing upon an earlier program of normative research into the links between a maturing sense of personal continuity and the development of a sense of commitment to the future, and capitalizing on recent methodologic advances in the study of young persons' maturing sense of self-continuity, a series of comparisons were made between 30 psychiatrically hospitalized adolescents, at varying degrees of risk to suicide, and a matched control group of their nonsuicidal agemates. The results of this study show that, while almost all of the hospitalized adolescents evidenced some degree of developmental immaturity in their attempts to reason about their own identity across time, the high-risk suicidal group was unique in their special inability to locate any grounds upon which to justify their own continuity through time. These findings are interpreted in terms of their relevance for understanding both the normal identity formation process, and for the diagnosis and treatment of adolescents at special risk to suicide.